Global Environmental Change 24 (2014) 193202

Contents lists available at ScienceDirect

Global Environmental Change

journal homepage: www.elsevier.com/locate/gloenvcha

Policy support, social capital, and farmers adaptation to

drought in China
Huang Chen, Jinxia Wang *, Jikun Huang
Center for Chinese Agricultural Policy, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, No. Jia 11, Datun Road,
Anwai, Beijing 100101, China

A R T I C L E I N F O A B S T R A C T

Article history: Increasingly severe drought has not only threatened food security but also resulted in massive socio-
Received 31 July 2012 economic losses. In the face of increasingly serious drought conditions, the question of how to mitigate
Received in revised form 1 November 2013 its impacts through appropriate measures has received great attention. The overall goal of this study is to
Accepted 13 November 2013
examine the inuence of policies and social capital on farmers decisions to adopt adaptation measures
against drought. The study is based on a large-scale household and village survey conducted in six
Keywords: provinces nationwide. The survey results show that 86% of rural households have taken adaptive
Drought
measures to protect crop production against drought, most of which are non-engineering measures. In
Adaptation measures
the case of non-engineering measures, changing agricultural production inputs and adjusting seeding or
Policy support
Social capital harvesting dates are two popular options. A multivariate regression analysis reveals that government
China policy support against drought such as releasing early warning information and post-disaster services,
technical assistance, nancial and physical supports have signicantly improved farmers ability to
adapt to drought. However, since only 5% of villages beneted from such supports, the government in
China still has signicant room to implement these assistances. Moreover, having a higher level of social
capital in a farm household signicantly increases their adaptation capacity against drought. Therefore,
the government should pay particular attention to the farming communities, and farmers within a
community who have a low level of social capital. Finally, farmers ability to adapt to drought is also
associated with the characteristics of their households and local communities. The results of this study
also have implications for national adaptation plans for agriculture under climate change in other
developing countries.
2013 Elsevier Ltd. All rights reserved.

1. Introduction
There is growing concern about global climate change. Over
the past 100 years, the average global surface temperature has
increased by 0.74 8C (IPCC, 2007). From 1961 to 2011, the annual
average surface temperature in China rose by 0.29 8C every 10
years (NCC of CMA, 2011). In addition, from 1961 to 2011, 6 of the
10 major river basins in China recorded a declining trend in total
surface water, indicating signicant shifts in precipitation
patterns. Of these six river basins, the total surface water of the
Haihe River, Yellow River and Liaohe River has been decreasing by
3.64%, 2.26%, and 1.66% per decade, respectively (NCC of CMA,
2011).
Drought is an increasing problem and becoming more severe in
many regions in the world, including China. It is forecasted that the
* Corresponding author. Tel.: +86 10 64889841; fax: +86 10 64856533.
E-mail address: jxwang.ccap@igsnrr.ac.cn (J. Wang).
total area suffering from drought in the world will expand by 15
44% from now until the end of 21st century (IPCC, 2012). In China,
from the 1950s to the beginning of this century, the annual
average crop area suffering from drought has expanded from 11.6
million hectares to 25.1 million hectares, an increase of 116%. Over
the same period, the proportion of crop area hit by drought
increased from 8% to 16% (NBSC, 2010; MWR, 2010). Moreover,
the share of seriously damaged area (a yield loss of at least 30%) to
drought-hit area (a yield loss of at least 10%) increased from 34% in
the 1950s to 46% in the 1990s, and 58% in the rst ten years of the
21st century.
Increasingly severe drought has not only threatened food
security but also resulted in massive socio-economic losses.
Drought is known to be one of the most severe natural disasters
threatening food security (UNDP, 2004). For example, as a result of
serious drought, the grain yield fell by 59% and grain exports
decreased by 19% in Australia in 2002 (Karoly et al., 2003). In 2010,
Russia encountered its most severe drought in 130 years and
declared a temporary ban on the export of wheat and other

0959-3780/$ see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.gloenvcha.2013.11.010
194 H. Chen et al. / Global Environmental Change 24 (2014) 193202
agricultural products, which is considered one of the major causes
of the global food price hikes in 2010 (Wegren, 2011; Dronin and
Kirilenko, 2011). Over the past two decades, drought has resulted
in an annual grain production loss of more than 27 million tons in
China. In 2000, drought caused the largest grain production loss
(60 million tons) and made it difcult for 28 million people and 22
million heads of livestock to obtain drinking water (MWR, 2010). In
2007 and 2009, drought resulted in direct economic losses of over
RMB 100 billion Yuan in China (MCA, 2010).
In the face of increasingly serious drought conditions, the
question of how to mitigate its impacts through appropriate
measures has received great attention. The international commu-
nity has called for national development plans to incorporate
climate change adaptation (Adger et al., 2007; World Bank, 2010).
In recent years, the Chinese government has also given top priority
to formulate and implement adaptation policies (NPC, 2011).
Indeed, a national plan responding to climate change was issued in
2007, which was followed by a publication of Chinas white paper
on national policies and actions against climate change in 2012
(NDRC, 2007, 2012).
However, the current level of knowledge is not sufcient to
support the implementation of Chinas national plan on adaptation
to drought and other extreme weather events. The major problem
is that most existing studies focusing on China are based on a
qualitative analysis. In addition, most of the publications are based
on local case studies, seldom drawing on a larger sample survey
covering more than one province. Despite the rich information
provided by the qualitative studies, it is difcult to provide robust
evidence to support the adaptation policies in China. For example,
Wang et al. (2013) developed an analytical framework to examine
climate change adaptation for saline agriculture in Jiangsu
province. Based on descriptive analysis, scholars have focused
either on farmers adaptation strategies in coping with drought in
certain regions (Liu et al., 2008; Ju et al., 2008; Sun et al., 2012; Su
et al., 2012; Sjogersten et al., 2013) or on macro-level adaptation
strategies to deal with the long-term effects of climate change (e.g.,
Pan and Zheng, 2010; Xia et al., 2008). Until now, according to our
knowledge, only two studies have applied quantitative approach to
analyze adaptation issues. Wang et al. (2008) applied a simulation
model to explore adaptation options to water scarcity in the Haihe
River Basin. Wang et al. (2010) applied an econometric model to
analyze farmers choices on crop structures and irrigation methods
in different climatic conditions.
With the rising signicance of drought, several questions need
to be answered to increase adaptation capacity. How have farmers
responded to drought? What major adaptation measures have
been adopted? Why are some farmers able to respond to drought
while others are not? Has any policy supported farmers when they
face serious drought? If yes, how effective have these policies been
in helping farmers to adopt adaptation measures? Then, as has
been found in other countries (Deressa et al., 2009; Katungi, 2007),
is adaptation closely related to farmers social capital in China?
Answering these questions is critical, not only to better understand
farmers responses to extreme weather events, but also to provide
empirical evidence for policy makers to help them formulate
adaptation plans and policies.
The overall goal of this study is to examine the inuence of
major factors, particularly policies and social capital, on farmers
decisions to adopt adaptation measures against drought in China.
To achieve this goal, we have the following two specic objectives.
The rst is to gain a better understanding of current adaptation
measures taken by farmers against drought. To do so, we
conducted a large-scale household and village survey in six
provinces across China. The second objective is to conduct both
descriptive and econometric analyses to identify and quantify the
inuence of policy support, farmers social capital, and other
factors on their decisions to adopt adaptation measures against
drought.
The rest of paper is organized as follows. In the next section, we
briey introduce the data used in this study. Section 3 discusses the
current situation in terms of adaptation measures used by farmers.
Section 4 presents the relationship between farmers responses to
drought and policy support, and between their responses and their
households social capital. Section 5 presents multivariate analyses
on the determinants of farmers adaptation. The nal section
concludes the paper.
2. Data
The data used in this study are collected from one large-scale
eld survey conducted in six provinces in China. These six
provinces are Hebei in the Haihe River Basin (RB), Jilin in the
Songliao RB, Anhui in the Huaihe RB, Sichuan in the Yangtze RB,
Yunnan in the Southwest RB, and Zhejiang in the Southeast RB (see
Fig. 1). When selecting provinces for the eld survey, we
considered differences in climate and water resources between
the northern and southern regions, as well as the diverse economic
development in these regions. For example, the survey samples
cover three river basins (Songliao, Haihe, and Huaihe RBs), which
are characterized by less precipitation and more frequent drought
conditions, especially in Haihe RB. During the period 19502000,
precipitation in these RBs declined by 50120 mm (ECSNCCA,
2011) and drought has become more serious (Wang et al., 2007). In
contrast, the other two river basins (the Yangtze and Southeast
RBs) have more abundant precipitation and water resources, and
have witnessed an increase in annual precipitation of 60130 mm
over the same period (ECSNCCA, 2011), although with no
signicant change in drought conditions (Wang et al., 2007).
These six regions also represent high (Zhejiang Province), middle
(Jilin and Hebei Provinces), and low (Anhui, Sichuan, and Yunnan
Provinces) levels of economic development (NBSC, 2010).
Stratied random sampling was used in each province to select
the study areas. First, we divided all counties in each province into
three quantiles according to the per capita annual net income of
rural residents in 2009. In each quantile, we randomly selected one
county for the survey. Then, we randomly selected two townships
in each county and three villages in each township for eld
surveys. As a result, the survey samples included 108 villages, 36
townships, and 18 counties across the six provinces.
In each village, we conducted two surveys, namely village and
household surveys. In the village survey, the main respondents
were the village leaders, such as the village party secretary, the
village head, and accountants. The questionnaire mainly covered
two major issues: (1) policy support for dealing with drought,
including whether the government provided early drought
warning and prevention information services, as well as technical,
nancial, and physical assistance to farmers; (2) physical and
socio-economic conditions, such as water supply reliability, soil
type, transportation infrastructure, economic development, and
personal characteristics of the village leaders. Table 1 provides the
descriptive statistics for the major information collected from the
village surveys.
Within each village, we randomly selected 10 farm households
and then selected those households whose crop production had
suffered as a result of drought in the past ve years to be our study
samples. Here, drought is measured based on by a farmers own
judgment or self-report. If a farmer said that his/her agricultural
production was seriously inuenced by drought, we dene this to
mean that drought occurred or that the farmer experienced a
drought. Specically, during our survey, we rst asked farmers if
their agricultural production had been affected by drought in the
past ve years (20062010). If a farmer answered Yes then we
 H. Chen et al. / Global Environmental Change 24 (2014) 193202 195

Fig. 1. Locations of six provinces and river basins in China.

asked questions to establish which year(s) was (were) relatively
normal, and the year in which they experienced the most serious
drought (i.e., their agricultural production was seriously hit by
drought). We recorded the answer to the latter question as the
most serious drought year. The household survey was conducted
through face-to-face interviews. The results showed that about
53% of households (569 of 1080 households from 91 villages) had
suffered from drought during the period 20062010. Therefore,
these 569 rural households were used as the nal samples for this
study. In the rest of this paper, both the descriptive statistical
analysis and econometric analysis refer to this, sample of 569
households from 91 villages. We then asked a series of questions
on whether or not each of these farmers applied any adaptation
measures in response to the drought. In addition, we collected
basic characteristics of the households and their farms (e.g., social
capital, cropping structure, farmland, family wealth, and char-
acteristics of the household head), as these may inuence farmers
behavior when adopting adaptation measures. The descriptive
statistics of the major information collected from the household
survey are also summarized in Table 1.

Table 1
Descriptive statistics of village and households characteristics.

Mean Std. Dev.

Villages (sample = 91):

Received early drought warning and prevention information from the governments (1 = Yes; 0 = No) 0.57 0.50
Received technical, nancial and physical supports against drought from government (1 = Yes; 0 = No) 0.05 0.23
Surface water reliability (ratio of number of years with reliable surface water supply over 5, the past ve years) 0.58 0.43
Groundwater reliability (ratio of number of years with reliable groundwater supply over 5, the past ve years) 0.14 0.34
Sandy soil (1 = Yes; 0 = No) 0.34 0.48
Loam soil (1 = Yes; 0 = No) 0.36 0.48
Clay soil (1 = Yes; 0 = No) 0.30 0.46
Distance from township government ofce (km) 6.79 5.58
Number of enterprises in the village 2.16 4.83
Age of village leaders (years) 47.57 7.18
Education of village leaders (years) 6.60 3.01

Households (samples = 569):

Received early drought warning and prevention information from the villages (1 = Yes; 0 = No) 0.38 0.49
Received technical, nancial and physical supports against drought from villages (1 = Yes; 0 = No) 0.09 0.29
Social capital (number of relatives in government within 3 generations) 0.25 0.57
Crop type (1 = Grain; 0 = Cash crop) 0.87 0.33
Ratio of hilly, slopes and terraces over total cultivate land 0.45 0.39
Wealth (value of durable consumption goods) (10,000 RMB) 7.01 9.86
Population (number) 3.75 1.55
Age of household head (years) 50.74 10.27
Source: Authors survey.
196 H. Chen et al. / Global Environmental Change 24 (2014) 193202
3. Adaptation measures against drought by farm households
There are many denitions of the adaptation options. Since we
collected information on how farmers have actually responded to
drought, our analysis is about adaptation practices rather than
potential adaptation options. As summarized by the IPCC report
released in 2007 (Adger et al., 2007), adaptation practices can be
differentiated along several dimensions, for example, by type of
action (physical, technological, investment, regulatory, market), by
actor (national or local government, international donors, private
sector, NGOs, local communities, individuals), by spatial scale
(local, regional, national), among others. Based on our pretests and
experience during the eld survey, we found that classifying the
adaptation options as either engineering (e.g., investment or
maintenance options) or non-engineering (e.g., technological,
regulatory, or market options) types, farmers and local policy
makers nd them easier to understand. Therefore, we chose to
dene our adaptation options by type of action. Based on our
eld survey, engineering measures predominantly comprise
constructing wells, building cisterns, excavating channels, and
updating pump equipment. Non-engineering measures are char-
acterized by the nature of their technical, institutional, and legal
aspects. Therefore, we found that non-engineering measures in our
study areas mainly included changing crop production inputs,
adjusting planting or/and harvesting time, adjusting crop irriga-
tion, changing cropping patterns, and purchasing crop insurance.
The adaptation measures adopted by farmers could be adopted
before a drought (anticipatory) or after a drought (reactive).
However, since we did not ask farmers precisely when (e.g., which
month) they adopted their drought adaptation measures, we have
no way of differentiating between these two kinds of adaptation
measures in this study.
The eld surveys showed that most households took some sorts
of action to protect crop production in response to drought. For
example, of the 569 households who faced drought during the
study period, 86% adopted an adaptation measure, most of which
were non-engineering measures (see Table 2, row 3). Only 10% of
the households applied both engineering and non-engineering
measures (row 6). This is understandable, as non-engineering
measures are more convenient, less costly and more easily
implemented in the short term.
For those farmers who were able to apply engineering measures
in the face of drought, further analyses indicate that they did so
both to increase the amount of irrigation water and, to improve the
reliability of the water used in irrigation. Of the households that
Fig. 2. Engineering adaptation measures adopted by farmers for resisting drought.
Table 2
Adaptation measures against drought for households suffering drought.
Number of Share of
households households (%)
Total samples 569 100
Without adaptations measures 82 14
With adaptation measures 487 86
(1) Only engineering measures 0 0
(2) Only non-engineering measures 432 76
(3) Both types of adaptation measures 55 10
Source: Authors survey.
applied engineering measures, 33% invested in wells to access
groundwater resources and 25% chose to build cisterns to collect
rainfall (see Fig. 2). In addition, 15% of these households purchased
pumps to draw water from a nearby river or lake for irrigation. In
summary, 73% (33 + 25 + 15) of the households that adopted
engineering measures did so to increase their irrigation water
supply capacity. The remaining 27% of households invested in
water saving facilities to improve water supply reliability and
water delivery efciency. This group was further divided between
those who invested in surface pipe or sprinkler irrigation facilities
to improve their drought adaptation ability (14%), and those who
made an effort to maintain canals to reduce canal water leakage
and water delivery losses (13%).
The eld surveys revealed that changing crop production
inputs and adjusting crop planting and/or harvesting time are
two major non-engineering measures adopted by farmers.
According to the eld survey results, when shocked by drought
the major non-engineering measures chosen by farmers were to
change the crop production inputs (e.g., seeds, fertilizer,
pesticides, and labor). The share of households changing crop
production inputs accounted for 35% (Fig. 3). Moreover, 24% of
households preferred to adjust when they planted or harvested
their crops to reduce or prevent loss caused by drought. To
mitigate the potential negative impacts of drought, some
households (18%) also chose to enhance irrigation intensity by
changing irrigation times or volumes. Changing crop varieties or
planting drought-resistant crops were other non-engineering
adaptation measures adopted (14% of households). Finally,
buying crop insurance against more frequent extreme weather
events has also become a new option for farmers to mitigate the
impacts of drought on their income. According to our eld survey
data, of those households who adopted non-engineering mea-
sures, 8% bought crop insurance.
 H. Chen et al. / Global Environmental Change 24 (2014) 193202 197

Fig. 3. Non-engineering adaptation measures taken by farmers for resisting drought.

4. Policy support, social capital, and farmers adaptation 57% received early drought warnings and prevention information
to drought from local governments (Table 3, row 3, column 2). This group was
divided as follows: 15% received the information before the
4.1. Policy support and farmers adaptation to drought drought occurred (row 4), 9%received the information during the
drought, and 33% received information both before and during the
In general, when local governments at different levels (e.g., drought (row 6).
provincial, county and township) assist farmers against drought, Only 5% of villages enjoyed technical, nancial, and physical
there are two major approaches. The rst approach is to provide policy support from their local government (see Table 3, last row),
farmers early warning and prevention information against which was a much lower percentage than those that received
drought, and the second is to provide farmers technical, nancial, drought warning and prevention information. That is, during a
or physical support. Early warning and prevention information drought, most villages could not get policy support from their local
released to farmers often occurs through farmers meetings government on technical, nancial, or physical aspects. Therefore
organized by village leaders or township ofcials, broadcasts farmers in most of these villages (95%) have to depend on their own
and other media managed by the local governments, texting experience, labor, materials, or funds to face the drought and
message to farmers cell phones, and issuing urgent disaster reduce the possible economic losses.
documents from the higher to the lower levels of government, and Does providing information and policy support play a positive
from there to the village community. Information provided before role in helping farmers to adopt adaptation measures against
a drought occurs, mainly it emphasizes how to prevent potential drought? As shown in Table 4, both information provision and
losses by telling farmers the possible duration and seriousness of policy support are positively related to farmers adaptations. For
the forecasted drought, and by reminding farmers that they should example, in the villages that received the early warning and
take appropriate adaptation measures (e.g., storing water or prevention information, on average, 88% of farmers adopted
adjusting crop planting patterns) to reduce losses. In addition to adaptation measures, which was higher than the corresponding
informing farmers of the characteristics and intensity of a drought, number in the villages that did not receive such information
information provided during a drought is mainly to help farmers (Table 4, rows 1 and 5, column 1). Even when information was only
use remedial measures to minimize losses (e.g., exploring other provided before or during a drought, the percentage of households
possible water resources, adjusting the input of crop production, or with adaptation measures (90% or 86%) was higher than those that
changing the cropping system). did not receive any information (81%, row 5, column 1). Similarly,
On the other hand, with regards to technical, nancial and when the local governments offered technical, nancial, or
physical policy support, local governments often help farmers physical support to farmers, 96% of households took adaptation
more directly. For example, local governments often send measures (row 6). If such policy support was not offered, the
extension staff to the elds to advise farmers on how to adopt
engineering or non-engineering measures (e.g., teaching farmers
Table 3
how to build or maintain their infrastructure, how to adopt water
Provision of early warning and prevention information, policy supports against
saving technology, and how to adjust production inputs). Financial drought.
support provides subsidies to farmers through drought relief funds
Number of Share of
and subsidized loans. The nancial subsidy can relax the nancial
villages villages (%)
constraints on farmers to help them to adopt adaptation measures
or can directly reduce their losses from the disaster. The local All villages 91 100
Villages not received information 39 43
governments also provide farmers with materials (e.g., water- Villages received information 52 57
saving facilities, drought-resistant seeds, fertilizer, pesticide or Before drought only 14 15
some other production inputs) to help their ght against drought. During drought only 8 9
The survey results demonstrate that policy support differs Both before and during drought 30 33
largely among villages, which provides good empirical data for us Villages not provided policy supports 86 95
to examine the relationship between policy support and farmers Villages provided policy supports 5 5
adaptation to drought. For example, among the studied villages, Source: Authors survey.
198 H. Chen et al. / Global Environmental Change 24 (2014) 193202

Table 4
Relationship between the provision of early warning and prevention information, policy supports and the adoption of adaptation measures by farmers.

Share of households with adaptation measures (%) Of which:

Only non-engineering measures (%) Both types of measures (%)

Villages received information 88** 74* 14***

Before drought only 90* 86* 4
During drought only 86 77 9**
Both before and during drought 89** 70 19***

Villages not received information 81 78 3

Villages provided policy supports 96** 59 37***
Villages not provided policy supports 85 78 7
Source: Authors survey.
Note: The superscripts are t statistical test results. For rows 1 to 4, their comparing base is row 5 (Villages not received information). For row 6, its comparing base is row 7
(Villages not provided policy supports).
*
Statistically signicant at 10%.
**
Statistically signicant at 5%.
***
Statistically signicant at 1%.

percentage of households taking adaptation measures dropped to
85% (last row). The above analysis implies that, if local govern-
ments provide early warning and prevention information or
directly give technical, nancial, or physical support to farmers, the
probability of farmers adopting adaptation measures is likely to
increase and the adaptation capacity of farmers against drought
risks is likely to be enhanced.
What kinds of adaptation measures are more likely to be taken
by farmers when information and policy support are available? As
shown in Table 4, when policy support is available, farmers are
more likely to adopt both engineering and non-engineering
measures. For example, if early drought warning and prevention
information to farmers is not provided to farmers in the villages,
only 3% of households adopted both types of adaptation measures
(engineering and non-engineering measures) (column 3), which
was signicantly lower than the 14% of farmers that did so in the
villages that did receive the information. In addition, when
technical, nancial, or physical policy support were offered by
the local governments to the villages, the share of households
adopting both types of measures reached 37%, more than ve times
higher than those villages that did not receive policy support (7%).
However, if we check the relationship between the policy support
and the adoption of non-engineering adaptation measures only, it
is difcult to nd a consistent story given the simple descriptive
statistical data (column 2).
4.2. Social capital and farmers response to drought
In recent years, the inuence of social capital on the adoption of
adaptation measures has captured the attention of scholars. Social
capital is an aggregate of actual or potential resources which are
linked to the possession of a durable network of more or less
institutionalized relationships of mutual acquaintances (Bourdieu,
1985). With the development of social capital theory in recent
years, some scholars have studied the relationship between
farmers social capital and their adoption of agricultural produc-
tion technologies (Wu and Pretty, 2004). These studies have
recognized that social capital can be treated as one channel for
acquiring information and helping farmers reduce necessary credit
when applying for a loan (Katungi, 2007). Another potential
advantage of social capital is to promote the adoption of new
technologies, since the externality of new technologies can be
internalized if farmers within the social network work together to
overcome difculties. As a result of its importance, social capital
has become the focus of studies on adopting adaptation measures.
For example, Deressa et al. (2009) found that social capital can
signicantly increase the adoption possibilities of adaptation
measures by African farmers.
Social capital can be measured in two ways. The rst is as a
network resource, and the second is a Guanxi resource (Zhang,
2003). In China, one familys Guanxi resource is closely related to
the number of relatives working in the government (Luo, 1997).
Based on a case study, Yan (1996) found that farmers who have
relatives working within the government nd it easier to solve
problems that occur in the villages (Yan, 1996). Park and Luo
(2001) found that Chinese rms can benet through their personal
connections with government ofcials. Considering the important
role of Guanxi in China, scholars analyzing Chinese issues usually
use Guanxi to measure social capital (Hwang, 1987; Yang, 1994;
Xin and Pearce, 1996; Lin and Si, 2010). Therefore, to study
whether social capital drives farmers to adopt measures in China,
we collected information on the number of relatives of a household
(within three generations) who work in the township government
or upper levels of government, which we then used as an indicator
for social capital.
According to our survey data, households with stronger social
capital are more likely to take adaptive measures. As shown in
Table 5, there is a clear, positive correlation between households
that adopt adaptive measures and the number of relatives that
household has working in the government. For example, house-
holds that adopted adaptation measures had an average of 0.27
relatives working in government, which was more than twice as
large as those without adaptation measures (0.13). Interestingly,
the effort made by farmers to adopt these measures was also
positively related to their social capital. Households that only
adopted non-engineering measures had an average of 0.25
relatives working in the government, while those households that
adopted both engineering and non-engineering measures had 0.47
relatives working in the government.
Table 5
Relationship between social capital and the adoption of adaptation measures by
farmers.
Average number of
households relatives
(within 3 generations)
worked in the government
Household without adaptation measures 0.13
Household with adaptation measures 0.27**
Of which:
Only non-engineering measures 0.25*
Both types of measures 0.47***
Source: Authors survey.
Note: The superscripts are t statistical test results. The comparing base is row 1
(Household without adaptation measures).
*
Statistically signicant at 10%.
**
Statistically signicant at 5%.
***
Statistically signicant at 1%.
 H. Chen et al. / Global Environmental Change 24 (2014) 193202
5. Econometric model and estimation results
5.1. Specication of econometric models
Since the descriptive statistical analysis did not control for the
inuence of other factors, it is difcult to separate the inuences
of early warning and prevention information, policy support, and
social capital on farmers adoption of adaptation measures to
cope with drought. In addition, the personal characteristics of
households, and the local socio-economic and physical condi-
tions in the villages are likely to inuence the adoption of
adaptive measures (Deressa et al., 2009; Hassan and Nhema-
chena, 2008). Therefore, to better quantify the inuences of
different factors on farmers decisions to adopt adaptation
measures, based on our survey, we specied the following two
econometric models:
Ri j a bP j @Si j g Hi j mV j uDk ei j (1)
M i j s ? P j rSi j #Hi j dV j Dk t i j (2)
In model (1), Rij indicates whether household i in village j adopt
adaptation measure(s). This is a dummy variable with a value of 1 if
a household adopts adaptation measures (either engineering or
non-engineering measures), and zero otherwise. Model (2) is
similar to model (1), but here the dependent variable is Mij, which
can take one of three values. If a household did not adopt any
adaptation measures, Mij is set to 1. If a household only adopted
non-engineering measures, Mij is set to 2, and if a household
adopted both engineering and non-engineering measures, Mij is set
to 3.
The independent variables are dened as the follows. Pj is a
vector with two policy related variables, the most interesting
variables, measured at the village level: (a) whether or not a
village received early warning and prevention information (or
information service) and (b) whether or not a village received
policy support. For (a), we use three dummy variables to measure
the information service was information only provided before a
drought (1 = yes; 0 = no); was information provided during a
drought (1 = yes; 0 = no); and was information provided both
before and during a drought (1 = yes; 0 = no). These three
variables are compared against the option of no information
being provided by the local government. For (b), there is one
dummy variable, which is set to 1 if a village received policy
support from the local government (e.g., technical, nancial, or
physical support), and zero otherwise. The variable Sji represents
the social capital status of a household, and is one of the
interesting variables to be examined in this paper. To measure this
variable, as dened in the previous section, we use the number of
relatives a household has working within the government (within
three generations).
In addition to policy and social capital variables, we include
control variables to represent socio-economic and physical
conditions for the households and villages, as well as a regional
dummy variable that may possibly affect the adoption of
adaptation measures by individual households. Here, Hij is a
vector of variables used to reect the socio-economic character-
istics of households, comprising the following: (1) crop type 1 for
grain and 0 for non-grain crops; (2) farmland topography
characteristics, measured as the ratio of hilly, sloped, and
terraced to total cultivated land; (3) wealth, measured by the
value of a households durable consumption goods (RMB10,000);
(4) family size; and (5) the age of the head of the household (in
years). In addition, Vj represents the physical and socio-economic
conditions of a village, including: (1) the reliability of the supply
of surface water resources, measured as the ratio of the number of
199
years with reliable surface water supply over 5, representing the
past ve years; (2) the reliability of the supply of groundwater
resources, measured as the ratio of the number of years with
reliable groundwater supply over 5; (3) soil types, which is either
loam (1 = yes; 0 = no) or clay soil (1 = yes; 0 = no), and compared
against the amount of sandy soil; (4) the economic development
of the village, measured by the number of enterprises in the
village; (5) the transportation infrastructure, measured by the
distance from the village to the township government ofce
(km); and (6) the characteristics of the village leaders, measured
by the age of the village leaders (in years). In addition, we include
a provincial dummy variable (Dk) to control for the inuences of
regional characteristics that do not change over time, but that
may affect the adoption of adaptation measures among the
provinces (k). In the model, a, b, @, g, m, u, s, 1 , r, d and w are
parameters to be estimated, and eij and tij are random error terms,
and are assumed to be subjected to independent identical
distribution.
5.2. Estimation approach
Given the nature of the dependent variables, we used different
estimation methods. As the dependent variable in model (1) is a
dummy variable, a logit model is used to run the regression
(Wooldridge, 2002). In model (2), the dependent variable measuring
the adoption of adaptation measures is made up of three discrete
values (1, 2, and 3), which represent three mutually exclusive and
independent adaptation options. Considering the nature of this
dependent variable, we apply the multinomial logit model (MNL) to
estimate model (2). The major advantage of the MNL model is that it
allows one to analyze the determinants of various choice possibili-
ties (Wooldridge, 2002). For example, the MNL model has been used
to analyze the choice of adaptation measures adopted by Ethiopian
farmers in Africa (Deressa et al., 2009).
For both the logit and MNL models, the estimated coefcients
reported in Table 6 reect the direction of the inuence of the
independent variables on the dependent variable. The magnitude
of the inuence cannot be indicated by the coefcients. Therefore,
based on the coefcient estimation, we computed the marginal
effect of the key independent variables in the two models, as
shown in Table 7.
5.3. Estimation results
The estimated results for models (1) and (2) show that the
models perform well. The likelihood-ratio statistics for the two
models are all signicant at the 1% level, and passed the Chi square
test (see Table 6). The pseudo R2 is 0.0969 for model (1) and 0.1302
for model (2) (Table 6, row 22). These values are high enough for a
multivariate analysis based on cross-sectional data. Furthermore,
the sign of many household and village level control variables are
consistent with our expectations, as well as being statistically
signicant. For example, in model (1), the sign of the crop type
coefcient is positive and statistically signicant. This implies that,
after keeping all other factors constant, when grain is shocked by a
drought, as compared to cash crops, farmers are more likely to take
adaptation measures, particularly non-engineering measures
(Table 6, row 6). The nding that grain is more sensitive to
drought than cash crops is also consistent with other researchers
ndings (Deng et al., 2006). The results also indicate that
adaptation measures are more likely to be adopted in non-plain
farmlands (such as hilly, slopes, and terraces) than plain farmland
because of their higher vulnerability (Table 6, row 7, columns 1 and
2). Some physical and socio-economic conditions in the villages are
also correlated with the farmers decisions to adopt adaptation
measures. For example, the coefcient for the groundwater
200 H. Chen et al. / Global Environmental Change 24 (2014) 193202

Table 6
Estimation results on the determinants of adopting adaptation measures by farmers.

Model (1): Logit model Model (2): MNL model (Control group: not
adopting adaptation measures)

Adopting adaptation Only non-engineering Both types of

measures measures measures

Villages received information

1 Before drought only 0.844 0.838 1.165
(1 = Yes; 0 = No) (1.627) (1.607) (1.267)
2 During drought only 0.776 0.787 1.231
(1 = Yes; 0 = No) (1.247) (1.257) (1.207)
3 Both before and during drought 0.823* 0.739* 1.873**
(1 = Yes; 0 = No) (1.928) (1.730) (2.410)
Villages provided policy supports
4 Providing the supports 1.967** 1.693** 2.651***
(1 = Yes; 0 = No) (2.454) (2.083) (3.005)

Farmers characteristics
5 Social capital (number of relatives within 3 generations) 0.650** 0.602* 0.935**
(2.010) (1.853) (2.435)
6 Crop type (1 = Grain; 0 = Cash crop) 0.720* 0.781** 0.339
(1.883) (2.018) (0.627)
7 Ratio of hilly, slopes and terraces over total cultivate land 0.635* 0.627* 0.656
(1.759) (1.727) (1.197)
8 Wealth (value of durable consumption) (10,000 RMB) 0.00747 0.00733 0.00909
(0.482) (0.470) (0.425)
9 Family size (population) 0.0684 0.0608 0.175
(0.832) (0.736) (1.298)
10 Age of household head (years) 0.0152 0.0164 0.00241
(1.069) (1.149) (0.106)
11 Education of household head (years) 0.0110 0.0115 0.00438
(0.247) (0.255) (0.0625)

Villages characteristics
12 Surface water reliability 0.0960 0.0539 0.507
(0.226) (0.126) (0.770)
13 Ground water reliability 1.514*** 1.488*** 2.041**
(2.825) (2.768) (2.400)
14 Loam soil (1 = yes; 0 = no) 0.185 0.199 0.0412
(0.544) (0.580) (0.0818)
15 Clay soil (1 = yes; 0 = no) 0.0272 0.0185 0.0995
(0.0699) (0.0473) (0.134)
16 Number of enterprises in villages 0.109 0.114 0.0503
(1.354) (1.411) (0.419)
17 Distance from township government ofce (km) 0.0471** 0.0497** 0.0272
(2.029) (2.113) (0.793)
18 Age of village leader (years) 0.00463 0.00246 0.0288
(0.202) (0.107) (0.769)
19 Province dummy Not reported Not reported Not reported
20 Constant 1.931 1.789 0.656
(1.306) (1.207) (0.273)
21 LR chi2 45.48 105.83
22 Pseudo R2 0.0969 0.1302

Note: (1) Absolute value of z statistic in parentheses; (2) LR chi2 value is calculated as LR chi2 (22) in model 1 and LR chi2 (44) in model; (3) Sample is 569.
*
Statistically signicant at 10%.
**
Statistically signicant at 5%.
***
Statistically signicant at 1%.

Table 7
Marginal effects of provision of early earning and prevention information, policy supports and social capital on the adoption of drought adaptation measures by farmers.

Model (1): Logit model Model (2): MNL model

Adopting adaptation Not adopting Only non-engineering Both types of

measures (Yes = 1; No = 0) measures (Y = 1) measures (Y = 2) measures (Y = 3)

Villages received information

Both before and during drought 0.089 0.092 0.031 0.061

Villages provided policy supports

Providing the supports 0.166 0.163 0.114 0.049

Social capital
Number of relatives within 3 generations 0.093 0.091 0.079 0.012

Note: Only presents the marginal effects of key variables that are statistically signicant.
 H. Chen et al. / Global Environmental Change 24 (2014) 193202
reliability variable is positive and signicant in all models (row 13),
demonstrating that, after controlling for the inuence of other
factors, and increasing reliability of the groundwater supply can
signicantly enhance farmers ability to adopt adaptation mea-
sures against drought. Moreover, the estimated results in Table 6
also show that the distance from the township is negatively
correlated with farmers ability to adopt adaptation measures (row
17), which may indicate the potential role of a rural infrastructure
in adopting drought adaptation measures.
More importantly, the estimation results reveal that providing
early drought warnings and prevention information signicantly
promote the adoption of adaptation measures by farmers. The
estimated coefcients of the variables that represent the provision
of information, both before and during a drought, are positive and
signicant in all models (Table 6, row 3). The results of model (1)
demonstrate that providing early drought warning and prevention
information, both before and during a drought, can play a
signicant role in helping farmers decide to adopt adaptation
measures when facing a drought (column 1). The results of model
(2) further show that the information service helps farmers to
adopt both non-engineering measures (column 2) and engineering
measures (column 3), as the estimated coefcient of the variable
(both before and during a drought) is much larger in column 3 than
column 2 (Table 6, row 3). According to the calculated results on
the marginal effect, if local governments provide information both
before and during a drought, the possibility of farmers adopting
adaptation measures can increase by 8.9% (Table 7, row 1, column
1). Specically, the possibility of adopting non-engineering
measures rises by 3.1% (column 3), which is lower than the rate
of adoption of the two types of measures together (6.1%, column 4).
However, our results also show that if information is provided
only before or during a drought, the inuence on farmers
adaptation behavior is not statistically signicant, although it is
still positive in all models (Table 6, rows 1 and 2). Therefore,
although policies on providing early warning and drought
prevention information can effectively encourage farmers to adopt
adaptation measures, when information is only provided before or
during the drought, the effectiveness of the information signi-
cantly decreases.
Furthermore, the estimate results show that offering technical,
nancial or physical support policies signicantly facilitates
farmers in adopting adaptation measures against drought. These
policies not only foster farmers application of non-engineering
measures, but also signicantly speed up the utilization of both
engineering and non-engineering measures. In the estimated
results of models (1) and (2), the coefcients of the technical,
nancial and physical policy support variables are all positive and
statistically signicant (Table 6, row 4). That is, after controlling for
the inuence of other variables, implementing technical, nancial,
or physical policy support signicantly increases the possibility
that farmers will adopt adaptation measures. Overall, the
possibility of adopting these measures can be increased by
16.6%; 11.4% for non-engineering measures alone and 4.9% for
both engineering and non-engineering measures together (Table 7,
row 2). Therefore, from a policy point of view, governments should
provide both early warning and prevention information to farmers
and direct technical, nancial, or physical support.
We also noted the importance of households social capital to
farmers decisions to adopt adaptation measures. Based on the
estimation results of both models (1) and (2), the coefcients of the
social capital variable are positive and statistically signicant in all
models, implying that there is a positive relationship between
social capital and the adoption of adaptation measures (Table 6,
row 5). In other words, if households have more relatives working
in the village, township or other upper level governments, their
extensive social network allows them better communication with
201
outside societies, as well as the possibility of receiving more help or
advice on drought resistance. Therefore, they are more likely to
adopt adaptation measures. If a household increases the number of
such relatives by 1, the possibility for adopting adaptation
measures increases by 9.3% (Table 7, row 3, column 1). In
particular, the possibility of only adopting non-engineering
measures increases by 7.9%.
6. Concluding remarks
Based on large-scale eld surveys conducted in six provinces in
China, this study examined crop farmers practices when faced by
drought and identied the major factors that affect farmers
decisions on whether or not to adopt adaptation measures against
drought. The results show that, when facing a drought, the
majority of farmers (86%) in China adopt non-engineering
measures, while only 10% of farmers adopted engineering
measures. In the case of non-engineering measures, changing
agricultural production inputs and adjusting seeding or harvesting
dates are two popular options. Some farmers also chose to enhance
irrigation intensity, change crop varieties, plant drought resistant
crops, or even to purchase crop insurance. Engineering measures
adopted by farmers include those that can either increase the
supply of irrigation water or improve the reliability or delivery
efciency of irrigation water.
Further analysis indicates that government policy support
related to drought resistance can play an important role in help
farmers to adapt measures. These supports include providing
farmers with early warning and prevention information against
drought and assistances in terms of direct technical, nancial, or
physical support. In addition, having a higher level of social capital
in a farm household signicantly increases their adaptation
capacity against drought. However, it is also worth noting that
our study focuses only on farmers adaptation capacity against
drought. We have not examined the impact of improving farmers
adaptation capacity on their crop production and income. The
effectiveness and impact of policy support and social capital on
farmers agricultural production and livelihoods after implement-
ing the different adaptation measures are important research
issues for future studies.
The results of this study have several potential policy
implications. First, there is signicant room for the government
in China to provide the early drought warning and prevention
information to local villages and farmers. For example, in our study
area, about two thirds of the rural village farmers have not received
this information service. Second, China may need to continue to
expand its policy supporting farmers in ghting drought, but it
may also be worth examining the costs and benets of the
government direct policy support further, for two reasons. First,
there is great potential to expand these activities in China, but this
could also be expensive. In the study area, in recent years, only 5%
of villages beneted from government assistance in terms of direct
technical, nancial, or physical support. While we did not have
data on the cost of implementing these policies, in the eld survey,
we were told that the major reason for lower coverage of these
supporting policies is budget constraints as direct policy support
could be expensive.
Third, improving farmers social capital is a way to help farmers
increase their ability to adopt adaptation measures during
droughts, and the government should pay particular attention to
the farming communities, and farmers within a community who
have a low level of social capital. While this study uses the number
of relatives a household has working in the government as a proxy
for social capital, the results should have general implications for
other measures (e.g., exchanges and communication among
farmers and farmers cooperatives in rural China), which may
202 H. Chen et al. / Global Environmental Change 24 (2014) 193202
raise farmers social capital. The purposes of such measures are to
transform the potential social capital into tangible social benets,
to increase the likelihood of adopting adaptation measures, and to
improve the capacity of farmers to cope with drought.
Lastly, we believe that the results of this study also have
implications for national adaptation plans for agriculture under
climate change in other developing countries. Directly providing
early disaster warning and prevention information to farmers,
particularly small-scale farmers, in many developing countries is
still not common. Given the rapid development of communication
technology, the wide spread use of cell phones in rural areas, and
the cost effectiveness of texting message to individual farmers,
disaster information and prevention services in developing
countries should be explored in more detail.
Acknowledgements
We acknowledge nancial support from the National Natural
Sciences Foundation of China (70925001, 71161140351), the Ministry
of Science and Technology (2012CB955700, 2010CB428406), Inter-
national Development Research Center (IDRC), World Bank, and
Australian Center for International Agricultural Research (ACIAR).
References
Adger, W.N., Agrawala, S., Mirza, M.M.Q., Conde, C., OBrien, K.L., Pulhin, J., Pulwarty,
R., Smit, B., Takahashi, K., 2007. Assessment of adaptation practices, options,
constraints and capacity. In: Climate Change 2007: Impacts, Adaptation and
Vulnerability. Contribution of Working Group II to the Fourth Assessment
Report of the Intergovernmental Panel on Climate Change. Cambridge Univer-
sity Press, London, pp. 717743.
Bourdieu, P., 1985. The forms of capital. In: Handbook of Theory and Research for
the Sociology of Education. Greenwood Press, New York, pp. 241258.
Karoly, D., Risbey, J., Reynolds, A., Braganza, K., 2003. Global warming contributes to
Australias worst drought. Australasian Science 24 (3) 1417.
Deng, K., Ju, H., Xiong, W., Yang, X., 2006. The impacts of climate change on
agriculture in China. Chinese Agricultural Science Bulletin 22 (5) 439441.
Deressa, T.T., Hassan, R.M., Ringler, C., Alemu, T., Yesuf, M., 2009. Determinants of
farmers choice of adaptation methods to climate change in the Nile Basin of
Ethiopia. Global Environmental Change 19 (2) 248255.
Dronin, N., Kirilenko, A., 2011. Climate change, food stress, and security in Russia.
Regional Environmental Change 11 (1) 167178.
ECSNCCA (Edition Committee of the Second National Climate Change Assessment),
2011. Second National Climate Change Assessment Report. Science Press,
Beijing.
Hassan, R., Nhemachena, C., 2008. Determinants of African farmers strategies for
adapting to climate change: multinomial choice analysis. African Journal of
Agricultural Resource Economics 2 (1) 83104.
Hwang, K., 1987. Face and favor: the Chinese power game. American Journal of
Sociology 92 (4) 944974.
IPCC (Intergovernmental Panel on Climate Change), 2007. Summary for policy-
makers. In: Climate Change 2007: Synthesis Report. Contribution of Working
Groups I, II and III to the Fourth Assessment Report of the Intergovernmental
Panel on Climate Change. Cambridge University Press, London.
IPCC (Intergovernmental Panel on Climate Change), 2012. Summary for policy-
makers. In: Managing the Risks of Extreme Events and Disasters to Advance
Climate Change Adaptation. A Special Report of Working Groups I and II of the
Intergovernmental Panel on Climate Change. Cambridge University Press,
London.
Ju, H., Conway, D., Li, Y., Harvey, A., Lin, E., Calsamiglia-Mendlewicz, S., 2008.
Adaptation Framework and Strategy Part 2: Application of the Adaptation
Framework: A Case Study of Ningxia, Northwest China. AEA Group, UK.
Katungi, E.M., 2007. Social capital and technology adoption on small farms: the case
of Banana production technology in Uganda. Ph.D. Thesis. Department of
Agricultural Economics, Extension, and Rural Development, University of Pre-
toria, South Africa.
Lin, J., Si, S.X., 2010. Can Guanxi be a problem? Contexts, ties, and some unfavorable
consequences of social capital in China. Asia Pacic Journal of Management 27
(3) 561581.
Liu, C., Golding, D., Gong, G., 2008. Farmers coping response to the low ows in the
lower Yellow River: a case study of temporal dimensions of vulnerability. Global
Environmental Change 18 (4) 543553.
Luo, Y.D., 1997. Guanxi: principles, philosophies, and implications. Human Systems
Management 16, 4351.
MCA (Ministry of Civil Affairs, P.R. China), 2010. China Civil Affairs Statistical
Yearbook 2010. Beijing, China Statistical Press.
MWR (Ministry of Water Resources, P.R. China), 2010. Bulletin of Flood and Drought
Disaster in China 2010. China Water Power Press, Beijing.
NBSC (National Bureau of Statistics of China, P.R. China), 2010. China Statistical
Yearbook 2010. China Statistical Press, Beijing.
NCC of CMA (National Climate Center, China Meteorological Administration, P.R.
China), 2011. Report of China on Climate Change Monitoring Released. China
Meteorological Administration, Beijing.
NDRC (The National Development and Reform Commission, P.R. China), 2012.
Chinas Policies and Actions for Addressing Climate Change. The National
Development and Reform Commission, Beijing.
NDRC (The National Development and Reform Commission, P.R. China), 2007. China
National Plan for Coping with Climate Change. The National Development and
Reform Commission, Beijing.
NPC (The National Peoples Congress, P.R. China), 2011. The Twelfth Five-Year Plan
for National Economic and Social Development of Peoples Republic of China.
Beijing, The National Peoples Congress.
Park, S.H., Luo, Y., 2001. Guanxi and organizational dynamics: organizational
networking in Chinese rms. Strategic Management Journal 22 (5) 455
477.
Pan, J., Zheng, Y., 2010. Analysis framework of climate change adaptation and
policy implications. Population, Resources and Environment in China 20 (10)
15.
Sjogersten, S., Atkin, C., Clarke, M.L., Mooney, S.J., Wu, B., West, H.M., 2013.
Responses to climate change and farming policies by rural communities in
northern China: a report on eld observation and farmers perception in dry
land north Shaanxi and Ningxia. Land Use Policy 32, 125133.
Su, Y., Xu, J., Wilkes, A., Lu, J., Li, Q., Fu, Y., Ma, X., Grumbine, R.E., 2012. Coping
with climate-induced water stresses through time and space in the
mountains of Southwest China. Regional Environmental Change 12 (4)
855866.
Sun, Y., Zhou, H., Zhang, L., Min, Q., Yin, W., 2012. Adapting to droughts in Yuanyang
Terrace of SW China: insight from disaster risk reduction. Mitigation and
Adaptation Strategies for Global Change 113, http://dx.doi.org/10.1007/
s11027-012-9386-2.
UNDP (United Nations Development Programme), 2004. Reducing Disaster Risk: A
Challenge for Development. A Global Report. UNDP, New York. Available at
www.undp.org/cpr/whats_new/rdr_english.pdf.
Wang, J., Huang, X., Zhong, T., Chen, Z., 2013. Climate change impacts and adapta-
tion for saline agriculture in north Jiangsu province, China. Environmental
Science and Policy 25, 8393.
Wang, J., Li, H., Xia, J., Ren, G., 2008. Water shortage and adaptation measures under
climate change: simulation analysis in the Haihe River Basin. Advances of
Climate Change Research 4 (6) 336341.
Wang, J., Mendelsohn, R., Dinar, A., Huang, J., 2010. How Chinese farmers change
crop choice to adapt to climate change? Climate Change Economics 1 (3) 167
185.
Wang, Z., Zhai, P., Wu, Y., 2007. Analysis on drought variation over 10 hydrological
regions in China during 19512005. Plateau Meteorology 4, 874880.
Wegren, S.K., 2011. Food security and Russias 2010 drought. Eurasian Geography
and Economics 52 (1) 140156.
Wooldridge, J.M., 2002. Econometric Analysis of Cross Section and Panel Data. MIT
Press, MA.
World Bank, 2010. Economics of Adaptation to Climate Change: Synthesis Report..
Available at http://climatechange.worldbank.org/sites/default/les/docu-
ments/EACCSynthesisReport.pdf.
Wu, B., Pretty, J., 2004. Social connectedness in marginal rural China: the case of
farmer innovation circles in Zhidan, North Shaanxi. Agriculture and Human
Values 21 (1) 8192.
Xia, J., Tanner, T., Ren, G., Cheng, X., Wang, J., Wang, Z., Yan, M., Liu, X., Holman, I.,
2008. Potential impacts of climate change on water resources in China: screen-
ing for adaptation and management. Advances of Climate Change Research 4 (4)
215219.
Xin, K.R., Pearce, J.L., 1996. Guanxi: connections as substitutes for formal institu-
tional support. Academy of Management Journal 39, 16411658.
Yan, Y.X., 1996. The culture of Guanxi in a North China village. China Journal 35, 1
25.
Yang, M.M., 1994. Gifts Favors and Banquets: The Art of Social Relationships in
China. Cornell University Press, New York.
Zhang, W.H., 2003. Social capital: theoretical discussion and empirical studies.
Sociological Studies 4, 2335.